The INCISIVE project has submitted and made public the ‘3.4 INCISIVE Standardization Suggestions’ deliverable, which includes recommendations for any worldwide project or initiative that wants to operate with global standards in the healthcare sector. The document also presents the interoperability framework that the consortium has developed to establish a federated repository of cancer medical images and clinical data that allows secure data sharing in compliance with ethical, legal and privacy requirements, and create an AI-based toolbox for incorporating AI features that improve cancer imaging and enable effective decision-making.

As described in the document, interoperability refers to the capacity of different systems, devices, or applications to exchange information without altering its meaning, and it is crucial to resolve data integration issues between systems, mitigate data inconsistencies, and enable effective management of potential changes.

Sara Alabart Martínez, INCISIVE’s interoperability expert from TIC Salut Social Foundation and leader author of the document, highlights that “in healthcare information systems, it is vital (among other aspects) to consider preserving the architecture, authorizing access, protecting and securing the data, maintaining meaning for the correct interpretation of the information by healthcare professionals, and guaranteeing focus on relevant patient information”.

The interoperability framework is the document that resolves and defines the process followed in a project and provides a complete overview of the standardization actions carried out. This includes the analysis of the technical and functional prerequisites, the actors and transactions of each use case involving the data use, the justification of the chosen standards, the methodology and design carried out, the interoperability risk analysis, the interoperability maintenance plan, the publication and reuse of the implementation guides and recommendations and future actions.

INCISIVE Standardization Actions

Section 4 of the deliverable details how INCISIVE has developed its interoperability framework and achieved a common data model for all data providers in the federated platform, in which clinical data is structured in HL7 FHIR messages and medical images in DICOM files. The data standardization has made it possible to have a common model for processing data from different sources of origin for model training in the federated nodes.

INCISIVE followed the recommendations of the Refined eHealth European Interoperability Framework (ReEIF) that defines all the aspects that must be considered to implement and involve all stakeholders in interoperability healthcare processes successfully. Then, the project specified the following interoperability layers:

• Legal interoperability (Legal and regulatory, Policy): ensures that all the actors operate under the same legal framework, policies, security, and strategies.
• Organizational interoperability (Care Process): ensures that all the actors and transactions of different use cases are standardized in a common and global definition of these scenarios.
• Semantic interoperability (Information): encodes data to maintain the meaning of all information exchanged.
• Syntactic interoperability (Information, Applications): defines how data exchange is to be carried out and the definition of its structure.
• Technical interoperability (Applications, IT Infrastructure): defines the technical language that will support syntactic interoperability and relates to interconnection, integration, accessibility, and presentation of data, among others.

For each layer of interoperability, there are different types of interoperability standards. One of the document’s recommendations is using the most appropriate standards for each functionality and combining them with each other to achieve the most complete standardized service.

All things considered, the specification proposed by INCISIVE combines the standards most suitably for the data it handles: SNOMED CT and LOINC to encode clinical and laboratory data, DICOM to standardize medical images, HL7 FHIR to standardize the exchange of clinical data encoded, HL7 CDA to standardize the exchange of clinical documents, and IHE to follow a common definition of clinical use cases and processes.

Standardization suggestions

Besides using the standards according to their functionality, the document offers multiple standardization suggestions regarding the following aspects:

• Use cases and healthcare process definition (topics discussed: IHE domains, ISOs, EMDN and MDC, OASIS, ATNA, IVDR, EUDAMED, etc.).
• Data exchange without losing the semantic meaning with terminologies (topics discussed: SNOMED CT, LOINC, RxNorm, ICD, NANDA, UCUM, CPT, MED-RT, NDF-RT, UNII, NDC, CVX, HPO, MEDCIN, ATC, etc.).
• Clinical documents exchange (topics discussed: HL7 CDA, FHIR Documents Module, FHIR XDS, IHE XDS, etc.).
• Medical image exchange (topics discussed: DICOM, DICOM SOP, DICOM SR, HL7 FHIR, etc.).
• Clinical data and events exchange (topics discussed: HL7 FHIR, RESTful API, HL7 v2, HL7 v3, etc.).
• Common data model to index or group data (topics discussed: OMOP, OMOP CDM, OHDSI, HL7 FHIR, ElasticSearch, Algolia, etc.).
• Standardize database with archetypes to present data in the user interface (topics discussed: OpenEHR).

INCISIVE HL7 FHIR implementation guide

The document includes a section about how to use an existing interoperability framework and an implementation guide. It also explains how to use the INCISIVE HL7 FHIR implementation guide published on Simplifier.net.

Relevant input for the EUCAIM project

Finally, this deliverable, together with the related input of all other AI4HI projects, is being used as one of the major contributions to the EUCAIM interoperability framework, which eventually may provide input to the European Health Data Space (EHDS) interoperability framework.

For more information, the ‘3.4 INCISIVE Standardization Suggestions’ deliverable is available in the open deliverables section of this website: https://incisive-project.eu/deliverables/